Right now we don't rank endpoints by latency, we just use the first one with the required consistency level. This is fine for local denokv server, but the cloud KV can provide multiple geo-located endpoints, and the spec recommends selecting based on latency:
From this list, the client MAY make requests to any endpoint, but the client SHOULD prefer endpoints that have lower latency.
Right now we don't rank endpoints by latency, we just use the first one with the required consistency level. This is fine for local denokv server, but the cloud KV can provide multiple geo-located endpoints, and the spec recommends selecting based on latency: